New (test) NTSC to PAL Script

[Xesdeeni]

I think my conversion script (http://forum.doom9.org/showthread.php?s=&threadid=45459&postid=537723) (see August 27 post) from PAL to NTSC works pretty well for an interlaced source. [Some would prefer slowing down the video for a progressive source, but I don't like processing the audio, so I prefer using a modified version (http://forum.doom9.org/showthread.php?s=&threadid=45459&postid=568758) (see the November 11 post) for progressive content. (I also proposed another trick (http://forum.doom9.org/showthread.php?s=&postid=603730) that might be even better.)] I prefer to use ChangeFPS() instead of ConvertFPS(), because the field blending is more annoying to me than the slight hitch on the repeated field.

But on PAL to NTSC, using ChangeFPS() will actually throw out fields. I live in NTSC (evolving to ATSC :)) land, so I can't vouch for this, but I think ConvertFPS() should look better for NTSC to PAL...except...due to the non-integer relationship of 59.94 fps to 50 fps, there are more blended frames than there are with hardware converters. In fact, there might be almost no unblended frames.

So I wanted to see if I could create a conversion that would minimize the blended frames. With that, I present the following test script. As I said, I live in NTSC land, so I can't really tell what this looks like in PAL. But if some kind soul(s) would like to give it a try and see what it does, I would be very grateful.

In the end, this probably needs to be done in some more efficient manner--perhaps a plugin. But this should suffice for test purposes:
xxxSource("NTSC.xxx")
AssumeFrameBased()
AssumeBFF() # adjust as necessary
KernelBob(order=0) # match order to above
AssumeFPS(60)
ab = SelectEvery(6,0,1,2,3,5)
cde = SelectEvery(6,0,1,2,4,5)
fgh = SelectEvery(6,0,1,3,4,5)
ij = SelectEvery(6,0,2,3,4,5)
abcde = ab.Overlay(cde, mode="blend", opacity=0.6)
fghij = ij.Overlay(fgh, mode="blend", opacity=0.6)
abcde.Overlay(fghij, mode="blend", opacity=0.5)
AssumeFPS(49.95)
ChangeFPS(50)
LanczosResize(720,576)
AssumeFrameBased()
AssumeTFF()
SeparateFields()
SelectEvery(4,0,3) # or SelectEvery(4,1,2) for BFF
Weave()

Xesdeeni

[scharfis_brain]

I've tested it.

It gave me uneven motion. So it will not be usable for me.
a plain convertfps(50) gave me better motion. I do prefer some more blends over heavy uneven motion.

but the best, we can use for 60i <-> 50i is mvbob() with mvconvertfps()

the results are INCREDIBLE.

[Xesdeeni]

Originally posted by scharfis_brain
I've tested it.

It gave me uneven motion. So it will not be usable for me.
a plain convertfps(50) gave me better motion. I do prefer some more blends over heavy uneven motion.

but the best, we can use for 60i <-> 50i is mvbob() with mvconvertfps()

the results are INCREDIBLE.Can you give me more details? What do you mean by "it gave me uneven motion?"

For clarity, ConvertFPS(50) will end up blending almost all the time. So an NTSC source like:

0 2 4 6 8 A C E G I K M O Q S U W Y ...
1 3 5 7 9 B D F H J L N P R T V X Z

would convert to something like:

0 2+3 4+5 7+8 9+A B+C E+F ...
1+2 3+4 5+6 8+9 A+B D+E F+G

With the above script, you should get something like:

0 2+3 4+5 7 9+A B+C E ...
1+2 3+4 6 8+9 A+B D F+G

The motion should be very similar, but there will be fewer blended frames.

Xesdeeni

[Xesdeeni]

P.S. The only MVTools I can find (http://jourdan.madism.org/~manao/) doesn't have MVBob().

Xesdeeni

[scharfis_brain]

The motion should be very similar, but there will be fewer blended frames.

This will not be possible!

I'll sort up the current blending techniques by motion smoothnes:
- Either do no blending (changefps) -> stutter
- do your approach -> less blending, but motion is pumping (sloooow - fassst - sloooow - fassst)
- do convertfps -> more blending but smoother motion, but moving borders are getting the infamous backwards rolling effect.
- do overlay(last,last.trim(1,0),opacity=0.5).convertfps() -> every frames gets blended, but motion is super smooth!

only the last both techniques are built into blendingmachines (standards converters)
I own a Sting Live-DVD, which has been converted using the last mentioned super-blender.
Its result is not bad.

Only motion compensated techniques are far better for converting 59.94i to 50i.
I have some raw scripts here, that do exactly this.
(mocomped deinterlace with mocomped framerate conversion and some other image enhancement)

mvbob() is a script cooked up by Mug Funky and overtaken by me (sorry Mug ;) )
http://forum.doom9.org/showthread.php?s=&threadid=84725

[Xesdeeni]

Originally posted by scharfis_brain
The motion should be very similar, but there will be fewer blended frames.

This will not be possible!

I'll sort up the current blending techniques by motion smoothnes:
- Either do no blending (changefps) -> stutter
- do your approach -> less blending, but motion is pumping (sloooow - fassst - sloooow - fassst)
- do convertfps -> more blending but smoother motion, but moving borders are getting the infamous backwards rolling effect.
- do overlay(last,last.trim(1,0),opacity=0.5).convertfps() -> every frames gets blended, but motion is super smooth!

only the last both techniques are built into blendingmachines (standards converters)I've been studying the output of standards converters for quite some time, and I've never seen one that produces blending on every frame. I've spoken at length with an engineer at Snell & Wilcox (and he gave me some documentation on their standards conversion--which is now available on their web site) on the subject. The outputs seem to be similar to the script above, which is precisely why I created it in the first place. Of course, I had to hand-calculate the ratios, and perhaps that is why you are seeing uneven movement. But that should be a simple matter of adjusting the coefficients. Any help you could provide in that area would be appreciated.

I'll have to digest the thread. Early on you reduce the height by 2 after KernelBob(). Is that always the case? Doesn't the reduce your resulting vertical resolution?

Xesdeeni

[scharfis_brain]

I've been studying the output of standards converters for quite some time, and I've never seen one that produces blending on every frame.

If I can convince my mum to hand out this DVD to me, I'll show you!

I've spoken at length with an engineer at Snell & Wilcox (and he gave me some documentation on their standards conversion--which is now available on their web site) on the subject. The outputs seem to be similar to the script above, which is precisely why I created it in the first place. Of course, I had to hand-calculate the ratios, and perhaps that is why you are seeing uneven movement. But that should be a simple matter of adjusting the coefficients. Any help you could provide in that area would be appreciated.

Can you set a link to tha paper?
Snell & Willcox are publishing a lot of papers.
I am sure, that this rule is fixed:
the more blending, the smoother the motion and vice versa.
But maybe you can get my opinion turned with some facts and sample scripts.
Also I recently had a clip to analyse, that had every field blended, too.
but the source was 23.976fps telcined film!
all fields of the resulting PAL-stream were blended.
some were normal blends, like you should know, and the others were NEGATIVE blends (the throw a inverted shadow!)
Many thanks to Didée for him creating a script that undoes this.

I'll have to digest the thread. Early on you reduce the height by 2 after KernelBob(). Is that always the case? Doesn't the reduce your resulting vertical resolution?

This is only the analysis chain.
It never reaches the output.
verticalreduceby2 is needed there to get more precise vectors.
I know, at first this sounds weird. But it returns more stable vectors. (bobbing jitter!)

[Xesdeeni]

Originally posted by scharfis_brain
I've been studying the output of standards converters for quite some time, and I've never seen one that produces blending on every frame.

If I can convince my mum to hand out this DVD to me, I'll show you!Are you sure they didn't use ConvertFPS()? ;)I've spoken at length with an engineer at Snell & Wilcox (and he gave me some documentation on their standards conversion--which is now available on their web site) on the subject. The outputs seem to be similar to the script above, which is precisely why I created it in the first place. Of course, I had to hand-calculate the ratios, and perhaps that is why you are seeing uneven movement. But that should be a simple matter of adjusting the coefficients. Any help you could provide in that area would be appreciated.

Can you set a link to tha paper?
Snell & Willcox are publishing a lot of papers.http://www.snellwilcox.com/knowledgecenter/books/books/estandard.pdf was one of them, and another was http://www.snellwilcox.com/knowledgecenter/books/books/emotion.pdf. But I don't see the one that was more specific to the four-field technique they actually use. Perhaps it was so old that it didn't make electronic publication?I am sure, that this rule is fixed:
the more blending, the smoother the motion and vice versa.I have never noticed this to be true. Since frames are discrete entities, blending them skews the timing (shows frames from different times). Any blending it noticable to me, and it never seems more smooth.But maybe you can get my opinion turned with some facts and sample scripts.I wouldn't want to attack your opinion. All of this is personal preference. I was just hoping to get some feedback on what the output of the above looked like on true PAL equipment. If you prefer more blending, then that is not necessarily wrong. I just prefer less. But reducing the number of blended frames doesn't help if I screw up the motion in the process :)Also I recently had a clip to analyse, that had every field blended, too.
but the source was 23.976fps telcined film!
all fields of the resulting PAL-stream were blended.
some were normal blends, like you should know, and the others were NEGATIVE blends (the throw a inverted shadow!)
Many thanks to Didée for him creating a script that undoes this.If there isn't a law against this, there should be :)

Xesdeeni

[scharfis_brain]

Are you sure they didn't use ConvertFPS()?

yes!
they used:
overlay(last,last.trim(1,0),opacity=0.5).convertfps()

to smooth out the motion.
at the cost that the motionblur of 1/60 sec became 1/30sec or more.

But I don't see the one that was more specific to the four-field technique they actually use.

I think, you are mixing up things:
two-field converter: stupid bobbing (linear interpolation)
four field converter: motion adaptive deinterlacing (comb mask)

those terms are not related to framerate conversion IMO.

Any blending it noticable to me, and it never seems more smooth.

to me blending is visible, too!
but please compare the motion smoothness yourself:

changefps() # no blending
convertfps() # partial plending
and
overlay(last,last.trim(1,0),opacity=0.5).convertfps() #total blending

there is a standalone frameblending program
http://converter.stratopoint.com/

that is capable of doing several kinds of blending. including thresholded blending (ie. blend weights under 20% (custom threshold) are getting ignored)
including the NEGATIVE blending I described in my last post.

I think, that such filter for AVS would be very cool.
(As standalone-app it is useless)

[Xesdeeni]

Originally posted by scharfis_brain
But I don't see the one that was more specific to the four-field technique they actually use.

I think, you are mixing up things:
two-field converter: stupid bobbing (linear interpolation)
four field converter: motion adaptive deinterlacing (comb mask)

those terms are not related to framerate conversion IMO.See the third section of this blurb: http://www.snellwilcox.com/applications/conversion.html.
Any blending is noticable to me, and it never seems more smooth.

to me blending is visible, too!
but please compare the motion smoothness yourself:

changefps() # no blending
convertfps() # partial plending
and
overlay(last,last.trim(1,0),opacity=0.5).convertfps() #total blendingEch! Three fields combined. What a blurry, ghosty mess!

I prefer the first for PAL to NTSC. But I think it throws out too much information for NTSC to PAL.there is a standalone frameblending program
http://converter.stratopoint.com/

that is capable of doing several kinds of blending. including thresholded blending (ie. blend weights under 20% (custom threshold) are getting ignored)
including the NEGATIVE blending I described in my last post.

I think, that such filter for AVS would be very cool.
(As standalone-app it is useless) Yes. That is kind of what the above script tries to do. If ConvertFPS() had a threshold, it might approximate this. Hmmm. Has anyone suggested this?

Xesdeeni

[Wilbert]

It gave me uneven motion. So it will not be usable for me.
a plain convertfps(50) gave me better motion. I do prefer some more blends over heavy uneven motion.

but the best, we can use for 60i <-> 50i is mvbob() with mvconvertfps()
I remember you said somewhere you are working on a new mvbob (perhaps you already made a new version). I'd appreciate it if you can add it to

http://www.avisynth.org/ShareFunctions

including a small readme how to use it and want plugins to install :)

[scharfis_brain]

@wilbert: it is far too early to publicate mvbob() in this way.

It hasn't reached a stable state jet.


Uhm, small documentation?!?
I prognose it becoming huge.

[scharfis_brain]

@xesdeeni:
If you are interested I could send some super-blended samples to you.

PM me for a transfer method of you choice.

btw. Did you know that I always need to type you nickname backwards? :D

I never can remember the letters in their 'normal' order ;)

[Xesdeeni]

Originally posted by scharfis_brain
@xesdeeni:
If you are interested I could send some super-blended samples to you.You mean using "overlay(last,last.trim(1,0),opacity=0.5).convertfps()"? I ran some tests with this. If the source is already blurred well on motion, it doesn't look too bad. It's a bit more blurry, but it doesn't have as much of the other artifacts. But if the source is less blurred on motion (like a higher speed camera or CGI), it really looks bad with as many as three frames shown at once.

I'm much more interested in the motion compensated work. I tried the first couple of scripts. My test clip had a bit better results than other methods in some cases. But your later, bigger, slower scripts were horrible. I'm not sure if my clip was pathological, or what. But the idea of using motion compensation on our lowly PC's is very exciting.btw. Did you know that I always need to type you nickname backwards? :DYou were the second to notice that in the 18 years I've been using it. But with this post, you won't be the last :)

Xesdeeni

[scharfis_brain]

But if the source is less blurred on motion (like a higher speed camera or CGI), it really looks bad with as many as three frames shown at once.

This is correct.
the super-blending only works with 1/60 sec shutter (NSTC) or 1/50 sec shutter (PAL) videos.

shorter shutter speeds will require motion compensation to avoid ghosting and motion judder.
everything else is unsatisfying.

But your later, bigger, slower scripts were horrible.

Maybe you didn't used the correct plugins.
My fault, cause I never wrote, which plugin versions to use with the current script.

To make you a little bit impatiently I want to say, that I nearly finished mvbob().

It seems to return really stable quality for now.

But it requires some packaging, like I did with R24 before, cause it is highly dependant on the correct versions of certain plugins.
Also it needs some documentation.

Altogether it will become a bundle, that allows you to do fully motion compensated standards conversion.
motion comensated deinterlacing with motion compensated framerate conversion :)

[manono]

Hi-

I've been studying the output of standards converters for quite some time, and I've never seen one that produces blending on every frame.

I have; quite a well known DVD also. It's the R3 EDKO HK DVD of House Of Flying Daggers (http://www.imdb.com/title/tt0385004/). It uses a very bad PAL master for the NTSC DVD. Fully 60% or so of the fields are blended, some more and some less, and every single frame is blended when there's movement of any kind. I took it, ran it through Restore24, and while it was impossible to remove all the blends, the improvement over the original is remarkable. In the process (also adding AssumeFPS(23.976) and adjusting the audio) I got rid of the PAL speedup, and restored the film to its original length. It's 114 minutes on the DVD, and was 119 minutes in the theater.

I respect your work with interlaced sources, Xesdeeni, but I think you're way off base recommending converting PAL 25p to NTSC 30i. Not only is interlacing all by itself harder to compress, but by encoding to 30fps, rather then to 24fps and running pulldown afterwards, you're encoding 25% more frames unnecessarily, resulting in a huge quality hit for the same file size. The same finite number of bits is then distributed among 25% more frames.

PAL(25p) -> NTSC(30i)

This closely resembles telecining film. Instead of a 2:3 pulldown, you get a 2:2:3:2:3 sequence. It jutters, but not much differently than telecined film, which NTSC viewers are used to. The only difference between this and the above conversion, is that deinterlacing is just field matching (in case the two fields you choose aren't in sync with the pair used for each frame), which is much less processor hungry.

and:

2. Change the speed to 23.976 fps progressive.

Pros: Progressive NTSC (easier to compress); no telecine artifacts on PC playback (multiple of 24 Hz refresh; otherwise, there are still telecine artifacts)
Cons: Audio manipulation (slowed with pitch change, or slowed with pitch adjustment artifacts); Program runs slower; Program lengthens

The Progressive age is upon us. As time goes by fewer and fewer of us will be using interlaced displays. Many, if not most, of the Progressive Scan DVD Players will screw up with interlaced encoding of what was originally 24fps FILM. They'll either just deinterlace it, or give you CUE, or both. Me, I'd much prefer to make the video progressive at almost all costs. AssumeFPS(23.976) is the only way to go for PAL to NTSC where the source was PAL 25fps progressive. Where a PAL master was used for the NTSC DVD, either RePAL (which scharfi doesn't much like) or Restore24, then followed by AssumeFPS(23.976) is a much better solution. Sometimes I think you make it more difficult than it really is. So, what's Con about adjusting the audio? A good AC3 encoder such as Soft Encode will do a fine job with it, put the pitch back where it was originally, and you'll wind up with a movie at its original and proper length.

This is a little bit off topic, but you referenced those older threads and posts earlier, and I thought I'd respond in this thread. Again, thousands have been helped by your scripts. I see them referenced all the time at Doom9 and at Videohelp.com. Even grumpy old neuron2 is always deferential to you, so you're obviously doing something right. :) Maybe because you come at this from your work with interlaced sources, you always (usually?) seem to take the interlaced approach. I come at this from years of encoding DVDs to AVI and now DVD, so I prefer a progressive approach. Plus, I hate blending with a passion, and can spot it a mile away. I'll do anything to get rid of it the best I can. I'm just an encoder, and not a developer by any stretch of the imagination, so I'll let you and scharfi get back to what you were talking about, and later on I hope to reap some of the rewards of the work the 2 of you do. Thanks for reading, and keep up the good work.

[Xesdeeni]

Originally posted by manono
Hi-

I've been studying the output of standards converters for quite some time, and I've never seen one that produces blending on every frame.

I have; quite a well known DVD also. It's the R3 EDKO HK DVD of House Of Flying Daggers (http://www.imdb.com/title/tt0385004/).It's well known that anime conversions are often horrible. As I said, I've never seen a professional standard converter output the crap I've seen in some anime conversions. I don't know what they use, but they are definitely not professional. :(I think you're way off base recommending converting PAL 25p to NTSC 30i. Not only is interlacing all by itself harder to compress, but by encoding to 30fps, rather then to 24fps and running pulldown afterwards, you're encoding 25% more frames unnecessarily, resulting in a huge quality hit for the same file size. The same finite number of bits is then distributed among 25% more frames.I haven't dismissed this. But the alternative is processing the audio. You can certainly choose to do this, but I have not been satisfied with the results, so I prefer the bit of additional video overhead, and that's what I recommend.The Progressive age is upon us. As time goes by fewer and fewer of us will be using interlaced displays. Many, if not most, of the Progressive Scan DVD Players will screw up with interlaced encoding of what was originally 24fps FILM. They'll either just deinterlace it, or give you CUE, or both. Me, I'd much prefer to make the video progressive at almost all costs. AssumeFPS(23.976) is the only way to go for PAL to NTSC where the source was PAL 25fps progressive.It is my understanding that the inability of players to handle deinterlacing of telecined material has to do with the incorrect flags in the stream. But YMMV.

I have been asking about whether you could use these flags on 25p video to create the best of both worlds, in another thread (http://forum.doom9.org/showthread.php?s=&threadid=88031). Maybe that would be a better compromise.So, what's Con about adjusting the audio? A good AC3 encoder such as Soft Encode will do a fine job with it, put the pitch back where it was originally, and you'll wind up with a movie at its original and proper length.I'm not sure a $500-$1000 tool is very useful to those of us just doing this as a hobby. For free, you have BeSweet or AVISynth. I haven't been impressed with the quality of speed adjusted audio with pitch compensation.Maybe because you come at this from your work with interlaced sources, you always (usually?) seem to take the interlaced approach. I come at this from years of encoding DVDs to AVI and now DVD, so I prefer a progressive approach. Plus, I hate blending with a passion, and can spot it a mile away. I'll do anything to get rid of it the best I can.'Tis true. I very rarely have a need to convert a movie. I typically convert either home video, or British Sitcoms. Obviously the former is interlaced. And the latter, unlike U.S. shows, are normally shot on video and are interlaced as well (except for exteriors, which are often 25p film, making the entire program a hybrid).

But my goal has always been reasonable quality with reasonable simplicity. One of the means is to limit the number of tools needed. And also, I try to limit the possible issues. Converting audio requires something to do the speed change/pitch compensation. Then audio encode is needed. While BeSweet is very nice, its AC3 encoder is not certified (in fact, I suspect BeSweet's AC3 encoder might be the reason my parents' DVD player cannot smoothly playback videos I give them on DVD--but that's for a different topic), and its user interface is horrid (I haven't tried out the new simpler GUI). Moreover, the chance of audio sync issues is multiplied.

Each conversion has its pros and cons. You may weight some differently than I do. I totally agree that blending sucks. That's where the above script came from. But I also weight distorted audio as a heavy con. And I'd rather have a slightly higher bit rate, or even risk an interlace artifact now and then. But you are free to choose otherwise.

If I ever get the time, I'll be updating my page. I intend to try to present as many methods as possible. But I still won't be able to present them all. And having a bunch of methods, will be more confusing for the one-shot converter. However, I think it's only fair to try to show some options.

Xesdeeni

[manono]

Hi-

It's well known that anime conversions are often horrible.

Except House Of Flying Daggers isn't anime. The director, Zhang Yimou, is very well known, with Ju Dou, Raise The Red Lantern, Happy Times, and Hero, among others, to his credit. Besides, but for some well known exceptions (Record Of Lodoss War and NGE, among others), the R1 anime DVDs are pretty good.

Some PAL to NTSC conversions are better than others. That House Of Flying Daggers was one of the worst I've seen. Recent R1 DVDs of BBC shows are better (but still not very good, since they blend also, just not as much). The Office, for example, was very watchable.

It is my understanding that the inability of players to handle deinterlacing of telecined material has to do with the incorrect flags in the stream.

I don't know if this is just a semantic difference or not, but once you encode already telecined 29.97fps (aka hard telecine), you are no longer encoding Progressive FILM, and you can no longer use RFF flags. Therefore the flags aren't set incorrectly, and pure flag reading players will spot the lack of such flags and deinterlace it.

The other thread to which you pointed was interesting. It is very possible to "telecine" 25fps material and put it on an NTSC DVD without blending. I've seen it only twice, though. Looking at the telecined frames, rather than 2 of 5 being interlaced, 2 of 6 were interlaced, and the fields were as you described them in that other thread. In fact, it was as a result of me sending neuron2 a vob sample of the R1 DVD of Johnny To's The Mission (http://www.imdb.com/title/tt0220644/) that he developed Guide=3 for Decomb's Telecide for 25->30fps type DVDs (not the blended kind, though). I've also seen a PAL DVD telecined just as you would NTSC Film. As a silent film buff, I see silents on DVD at all kinds of framerates. One such DVD, the R2 Pandora's Box (http://www.imdb.com/title/tt0018737/) (Die Büchse der Pandora) is on the DVD at 20fps, and had standard 2:3 pulldown applied to reach 25fps. Sadly, this isn't common practice, as many of the Chaplin R2 PAL short film DVDs, also 20fps, are field blended to reach 25fps. Another good restoration job ruined.

I haven't been impressed with the quality of speed adjusted audio with pitch compensation.

I agree that the BeSweet AC3 encoder isn't very good.

I suspect BeSweet's AC3 encoder might be the reason my parents' DVD player cannot smoothly playback videos I give them on DVD

Very possible, I think. You might try and run them through AC3Fix (http://www.digital-digest.com/dvd/downloads/showcategory.php?cid=5) next time to see if that helps.

[scharfis_brain]

@Xesdeeni:

Did you have some time to test my motion compensated standards converter script?

[Xesdeeni]

Originally posted by scharfis_brain
Did you have some time to test my motion compensated standards converter script?Have you updated it in the last day or so? (If so, I missed it, can you link to the thread, which I apparently didn't subscribe to?)

Xesdeeni

[scharfis_brain]

http://forum.doom9.org/showthread.php?s=&postid=606487#post606487

[kle500]

@scharfis_brain:

I tested your motion compensated standards converter script, and i was blown away with the quality of motion.
I am unable to see any blended frames, but i have a problem when there is scene change. I tested on a DVD m2v NTSC Interlaced clip
and when the frame change to another frame, i see many mini blocks (i think they are some kind of mask analysis).
I can send you 2 pics i captured from the avs script.
Tell me where i can send them.

As for speed........ It is Slooooow, but nothing like some other guy said (40sec clip in 21 hours).
For me, i converted a 4:30 min. clip and it took about 3:30 hours on P4 3GHZ.
God this is the best conversion i have ever seen.......
I will try to find some old DVD's (INTERLACED) i have, where there is fast camera panning, and in the past i remember that i was wishing to find a better way to handle those movements.
I also preffer the convertFps way with the blended frames, but smooth movement, rather than changeFps (stutter movement).

I wish you could make it faster.

@Xesdeeni:

Your script my friend works good for me.
There is no uneven motion to the clips i converted.
I will make a test with your original script and this new one, to see if there is any difference.
As you remember i still use the smoothdeinterlacer with the modifyed threshold settings, cause kernelbob, gives me thin black lines in fades.

I am in PAL land and i allways convert NTSC(i) -> PAL(i)

Keep up the good work people.

[Xesdeeni]

Originally posted by kle500
Your script my friend works good for me.
There is no uneven motion to the clips i converted.
I will make a test with your original script and this new one, to see if there is any difference.
As you remember i still use the smoothdeinterlacer with the modifyed threshold settings, cause kernelbob, gives me thin black lines in fades.

I am in PAL land and i allways convert NTSC(i) -> PAL(i)Hopefully, what you should see is 3 of 5 blended frames with the new script, while the old script will have almost all blended.

Xesdeeni

[kle500]

Yes, i see an improovement in blended frames with this new script.
I tested today and it is better than the original.
Mvbob on the other hand is Excelent in movement, allthough it is realy slow

[Kortik]

is there a step by step guide on how to convert NTSC DVD to PAL DVD?